The Pharmaceutical Journal Vol 263 No 7063 p436-446
September 18, 1999 The Conference

Microbiology: resistance, disinfection and adhesion

The session began, under the chairmanship of Dr G. W. Hanlon (University of Brighton) with a keynote lecture by A. D. Russell (University of Wales, Cardiff), which drew attention to an emerging problem, namely that resistance to bactericides is spreading in much the same way as has resistance to antibiotics.
Antibiotics usually work by inhibiting a specific biosynthetic process, whereas the attack by biocides is more general, and can have multiple target sites. Nevertheless, acquired resistance to inorganic and some organic mercurials is found on penicillinase plasmids in staphylococci, and Gram-negative bacteria that have developed resistance to cationic biocides, probably by multiple mutation, can show cross-resistance to some antibiotics.
This is believed to be due to outer membrane alteration, although efflux mechanisms (the pumping out of drugs by specific membrane proteins, a major contributory factor in antibiotic resistance) cannot at present be ruled out.
It has been proposed that the introduction of, for example, chlorhexidine, has been responsible for the selection of antibiotic-resistant strains of staphylococci and multidrug resistant Gram-negative bacilli, and recently it has been claimed that the bisphenol — triclosan — was selecting, in E coli, for chromosome-mediated resistance to antibiotics.
Resistance in Pseudomonas The first presentation exemplified the problem raised in the keynote lecture, being an examination of the development of resistance in Pseudomonas aeruginosa consequent upon repeated exposure to sub-lethal concentrations of benzalkonium chloride.
B. Forbes (King's College London), R. Lambert and J. A. Joynson (Unilever Research, Colworth) trained the organism to be resistant to the bactericide by repeated subculturing into gradually increasing concentrations. After 17 subcultures, the organism could survive at a level of 450 ppm, an increase of ninefold on the minimum inhibitory concentration (MIC) of the original, although the doubling time of the resistant strain was greater than that of the untrained strain.
Subculture in bactericide-free nutrient broth did not bring about a complete reversion to normal resistance levels, even after 34 subculture steps. No increase in resistance to tobramycin or amikacin could be detected, the MIC for all cultures remaining at 1-5 ppm. However, the results do show that exposure to disinfectant residues may well produce resistant strains under field conditions.
Staph aureus resistance The same problem was studied from the opposite end by M. T. E. Suller and A. D. Russell (University of Cardiff), who took some Staph aureus strains that were known to be resistant to the antibiotic methicillin, and others known to be sensitive. Similarly, they took some enterococci that were resistant, and others that were sensitive, to vancomycin. They tested the MICs of a range of bactericides against all of these strains. The bactericides were chlorhexidine, cetylpyridinium chloride, benzalkonium chloride, triclosan and dibromopropamidine isethionate.
The enterococcal strains showed no differences in resistance to any of the bactericides, irrespective of their antibiotic sensitivity, but things were different with Staph aureus. Antibiotic-resistant strains were up to three-fold more resistant to chlorhexidine than were the sensitive strains, up to four-fold more resistant to the cetylpyridinium and benzalkonium chlorides, more than 40-fold for triclosan, and 10-fold for dibromopropamidine isethionate.
The authors then tried to select for increased resistance, being successful to some extent, but finding that resistance induced by a stepwise procedure turned out to be unstable in the longer term.
All the MICs in the study are, of course, far lower than would be used in disinfection for clinical purposes.
Validation of disinfection techniques Aseptic dispensing units must provide products with assured quality, although, in the nature of the case, only limited, or no, final testing can be carried out, so that total reliance must be placed in the (validated) systems of work.
Official guidelines are not comprehensive, and the NHS quality control committee suggests that in-house standards should be set up where no sufficient standards have been officially defined. S. J. Hiom (St Mary's Hospital, Penarth, Glamorgan) described the validation of the disinfection techniques used during the transfer of small items (ampoules) from preparation areas to the work zone, using both the roll plate method and the measurement of total bioburden to quantify the degree of sterility obtained.
Ampoules of water for injections were exposed to natural contamination in a busy hospital corridor for 48 hours, then collected using sterile gloves and divided into batches for treatment by one of four disinfection procedures during transfer into an isolator.
The transferred ampoules were then tested by the roll plate method (5 seconds contact with a Biotest contact slide that was then incubated) and by the total bioburden method (sonication of the ampoule in sterile water, followed by plating out from the water and incubation, to determine the number of colony-forming organisms). The bioburden method turned out to be some six times more sensitive than the roll plate method at detecting surface contamination.
Of the disinfection techniques, in increasing order of effectiveness they were: spray with 70 per cent industrial methylated spirit (IMS) plus two-minute lock-out; spray with IMS with no lock-out; wipe with 70 per cent isopropyl alcohol (IPA); spray with 70 per cent IMS, plus wipe with IPA. All of these reduced contamination by a factor of 10 or more compared with untreated controls, but only the last came close to achieving complete sterility.
Post antibiotic effect Cells that have not died after exposure to an antibiotic do not grow so rapidly as they did originally when transferred into an antibiotic-free nutrient: this is called the post-antibiotic effect. It has been examined in E coli strain AB1157 by H. J. Wickens, R. J. Pinney (School of Pharmacy, University of London), V. A. Gant (University College Hospital) and D. J. Mason (University of Warwick).
The organism was cultured in nutrient broth, and then subcultured at a 1:50 dilution into broth containing four times the MIC of ciprofloxacin. After three hours there were 2 per cent survivors. These were washed free of antibiotic and resuspended in antibiotic-free nutrient broth. Viability as a function of time was determined by plating onto nutrient agar.
Loss of cell membrane integrity was measured by the uptake of the fluorochrome, propidium iodide, and loss of membrane potential by the uptake of another fluorochrome, bis-1,3-dibutylbarbituric acid trimethine oxonol.
After the ciprofloxacin treatment, 80 per cent of the cells were classified as filamentous, and their number remained constant for the duration of the post-antibiotic effect, which was 57 minutes under the conditions used. The increase in viable count that occurred at the end of this time was due to division of non-filamentous cells, and only 4 per cent of these were capable of forming colonies.
By studying the cultures using, simultaneously, their forward light-scattering and their fluorescence, it was established that there were two populations of cells present — fluorescent filaments and non-fluorescent rods. It appears that neither fluorochrome uptake nor filamentation are reliable indicators of cell death, since cells seem to enter a viable but non-culturable state, from which they may or may not emerge by intracellular repair.
Reducing bacterial adhesion Bacteria have the ability to protect themselves by forming adherent coatings on solid objects. Their ability to adhere to the surfaces of urinary catheters is one of the major causes of hospital infections, so J. Djokic, D. S. Jones, S. P. Gorman and S. McGrath (Queen's University of Belfast) have investigated the effects of including a bactericide in the polymer from which the catheters are made, to see whether such a modification would reduce the extent of adherence.
E coli, cultured from a retrieved urethral stent, were incubated with cast polycaprolactone films, containing 1 per cent of povidone iodine, in an orbital shaker at 37C for periods of up to eight hours. At regular intervals, the films were removed from the culture medium, washed and sonicated for 15 minutes to remove the firmly attached E coli cells, which were then tested for their viable count by serial dilution and plating out. Initially, there was no difference in adhesion of the organisms to the treated or the untreated films but, after two hours, the films containing povidone iodine resisted further adherence, so that by eight hours, the difference in adhesion was very pronounced, there being around 100 times more organisms per unit area on the untreated film.
Adhesion to fibrinogen Any implanted material coming into contact with blood is rapidly covered with a proteinaceous layer, usually of fibrinogen, one of the largest proteins found in normal plasma. Such films may then be more slowly invested by bacteria, many of which have specific receptors on their surfaces for host proteins to facilitate adhesion. There have been studies of the binding of S aureus to fibrinogen, but little attention has been given to the interaction of other pathogens with this protein. Accordingly, J. Martin, R. G. A. Faragher, G. W. Hanlon, A. W. Lloyd (University of Brighton) and S. Cederholm-Williams (Oxford Bioresearch) tested the degree to which a range of 19 pathogenic organisms were able to adhere to fibrinogen.
The method used was to prepare a suspension of polystyrene microspheres, 0.48mm diameter, coated with fibrinogen, in phosphate-buffered saline, together with a similar suspension of non-coated spheres. A drop of suspension was placed on a microscope slide and approximately five bacterial colonies from a pure culture on tryptone soya agar were added and mixed with the suspension using a sterile loop. The suspension was then observed and rated for its resulting degree of agglutination. Some organisms were able to bring about agglutination in the absence of a fibrinogen coating but several agglutinated as well, or better, when the coating was present. These included (besides Staph aureus) Staph haemolyticus, Strep faecalis, Proteus mirabilis, Edwardsiella tarda and Citrobacter freundii.
Surprisingly, since it is associated with many infections arising from indwelling devices, Staph epidermidis does not bind to fibrinogen, and so must colonise by a different route.

Neuropharmacology: Parkinson's disease, HRT and thyrotoxicosis

This session, running in parallel with that on controlled release, was chaired by Professor P. J. Nicholls (University of Wales , Cardiff).
The keynote lecturer was Professor N. G. Bowery (University of Birmingham medical school), who spoke about the current state of knowledge in the field of GABA receptors, of which, he said, there are two kinds — A and B. The A receptors are associated with fast synaptic events, while the B type produce slower but more prolonged changes in neuronal conductance. They are located both pre- and post-synaptically and on activation reduce Ca++ conductance or increase K+ conductance.
The receptor's structure has recently been determined — it exists as a heterodimer, which is unique amongst the metabotropic G protein-linked receptors. At present, only one drug, the antispasticity agent baclofen, is clearly known to act through it. This drug reduces the evoked release of excitatory transmitter from nerve terminals on to motoneurones in the spinal cord. Its action can also be analgesic, as intrathecal administration is effective in treating pain arising from head injury. However, local administration is necessary, to reduce side effects and because systemic administration rapidly leads to tolerance and lack of efficacy.
The mechanism of the analgesia appears to be inhibition of primary afferent transmitters, mediated by B receptors on the nerve terminals of small diameter fibres on the dorsal horn. Animal studies have indicated that B-receptor antagonists may have a number of therapeutic uses, including anti-epilepsy, cognition enhancement and antidepressant activity.
Common site for anaesthetic action General anaesthetics at clinically relevant concentrations have profound neurological effects, inhibiting both synaptic transmission and excitatory responses. Since the members of the nicotinic acetylcholine-like receptor super-family, which includes GABA, glycine and 5-HT3, are sensitive to anaesthetics, it would be interesting to know whether there is a common site within the receptor protein at which anaesthetics act.
To this end, S. Daniels, C. S. Fraser (University of Wales, Cardiff) and S. Wittmann (University of Regensburg, Germany) examined the effect of two structurally different anaesthetics — propofol and pentobarbitone — at the human homomeric a-1 glycine receptor. The cDNA that encodes the human glycine receptor was extracted from its residence plasmid, amplified, and transcribed to mRNA that was then microinjected into frog oocytes. These were freed from follicular and thecal cells, perfused with frog Ringer's solution and tested for glycine receptor function using a two-electrode voltage clamp technique.
Dose-response relationships were established for the bath application of each of the anaesthetics when the oocytes were stimulated with glycine. A series of further experiments, using various application regimes of the anaesthetics in combination, showed that propofol and pentobarbitone do indeed act at a common site, but that pentobarbitone acts additionally at a different site, at which it produces a glycine mimetic effect.
Direct action of levodopa Levodopa is the mainstay of the treatment of Parkinson's disease, with the accepted mode of action being its intraneuronal conversion to dopamine, producing release of the active transmitter from nigrostriatal terminals. However, recent data have indicated that there may well be a direct action of levodopa itself on dopamine receptors, bypassing the conversion to dopamine that is brought about by l-amino-acid decarboxylase.
To test this, C. S. Biggs, A. Fisher, O. Erdiri and M. Starr (School of Pharmacy, University of London) ran three parallel series of experiments to demonstrate the weakness of the link between levodopa-mediated restoration of movement in rats with induced Parkinson's disease, and the extent of synthesis and release of cerebral dopamine.
Male wistar rats were treated with reserpine, then divided into three groups. In the first group, animals were assessed for dopamine output in two regions of the brain — the substantia nigra and the contralateral striatum — by in vivo probe microdialysis at 15 minute intervals. Benserazide and a-methyl-p-tyrosine were administered at 2 hours, followed by levodopa at various dosages from 0 to 200mg/kg, dopamine content being assayed throughout. Animals in the second and third groups were similarly dosed.
Those in the second group were sacrificed for brain dissection, with analysis of the substantia nigra and contralateral striatum for l-amino-acid decarboxylase content.
The third group had the comparative good fortune to be placed in test areas where their motor activity was assessed by microwave doppler counting. The overall result was that the restoration of motor activity by levodopa does not correlate with nigrostriatal neuronal activity.
This mismatch is to be further investigated by trying to find the source of the dopamine recovered in the microdialysis, as this also does not correlate with motor activity.
Kinetics of apomorphine In humans, parkinsonism is not simple to control, since some patients under a levodopa regime can develop an on-off response, in which their clinical state fluctuates abruptly between periods of relative mobility and periods of severe parkinsonian disability, despite optimally timed doses of levodopa.
W. M. Ingram (University of Plymouth), M. J. Priston (Derriford Hospital, Plymouth) and G. J. Sewell (University of Bath) have examined the pharmacokinetics of apomorphine, a dopamine receptor agonist, which is used in anti-parkinsonian drug regimes to reverse refractory motor fluctuations.
There is a large inter-patient variation in drug absorption, indicating a need for individual handling of the administration of the drug. The authors developed a reversed-phase HPLC method for its estimation to help with pharmacokinetic measurements of plasma levels, which they endeavoured to correlate with the results of tapping tests. The work is in its early stages, and with only five patients in the test group it is difficult to quantify the response to apomorphine.
Vasodilatation by tamoxifen and oestradiol It is known that oestrogens in hormone replacement therapy (HRT) give a degree of protection against coronary disease, and tamoxifen has been reported to reduce the risk of myocardial infarction in postmenopausal women. Both of these effects may be due to vasodilator action.
J. R. McCurrie and I. Strati (University of Bradford), knowing from earlier Bradford work that tamoxifen relaxes rat aorta, have examined its effect on rat portal vein and compared it with the effect of 17-b-oestradiol.
Rat portal veins were set up in Krebs solution containing 10 mM indomethacin, and a control concentration-response curve was established for the effect of potassium chloride, over the range 10-80 mM. This experiment was then repeated after incubation with oestradiol or tamoxifen at 4, 10 or 20mM concentrations.
Both substances reduced the effect of the potassium chloride in a concentration-related manner, oestradiol by up to 60 per cent and tamoxifen by 40 per cent. However, a combination of both compounds at the 4 mM level relaxed the potassium chloride mediated contraction to a greater than additive extent, and, even more oddly, this synergism was reversed at the higher concentrations, where the combined effect was less than the additive value.
Further experiments, involving the combined effects of calcium and potassium ions, and the use of verapamil (a calcium channel blocker) showed that, although both oestradiol and tamoxifen relax vascular muscle, differences in their actions on calcium and potassium-induced contraction imply that yet a third mechanism must be involved, such as an inhibition of protein kinase C.
HRT benefits in old age J. R. McCurrie, this time with P. Eseh-Sumbele (also University of Bradford) reported on an associated investigation intended to find out whether the beneficial effects of HRT would be maintained in older age groups.
To this end, they tested the effect of 17-b-oestradiol on the contractile actions of angiotensin II, prostaglandin F-2a and potassium chloride on both the aorta and the portal vein of young adult rats (three months old) and extremely aged ones (24 months old).
Preparations were set up as in the previous communication. The development of tension was the dependent variable in the aortal testing, where no age effect was displayed in the responses to oestradiol or prostaglandin, but where the angiotensin response was only 25 per cent as large in the old rats. In the portal vein tests, where the response was relaxation, there was no age effect at all.
The good news seems then to be that, if humans respond in the same way as do rats, the beneficial effects of HRT should continue into old age.
Verapamil in thyrotoxicosis Verapamil, the calcium channel blocker used in the relaxation studies just described, was also used by P. A. Bhatt, M. Desai and D. D. Santani (Ahmedabad, Gujarat, India) in a study of its effects in thyrotoxicosis.
The thyroid hormones maintain metabolic homoeostasis, affecting most organ systems and are crucial determinants of normal development. In particular, calcium levels in the sarcolemma and the sarcoplasmic reticulum are affected by alterations in thyroid function.
The authors induced hyperthyroidism in guinea pigs by administration of l-thyroxine over a two-week period, with a group of the animals then being dosed for a further week with verapamil. Body weight, ECG, time required to onset of histamine-induced asphyxia, force of heart contraction and sensitivity of smooth muscle to histamine and 5-HT were all measured.
Verapamil controlled sinus tachycardia, atrial fibrillation, cardiac hypertrophy and asphyxia onset time in the treated animals, indicating that there is a significant role for calcium ions in the pathophysiology of thyrotoxicosis.

Short papers on controlled release

This first session started on the Monday afternoon, under the chairmanship of Professor H. N. E. Stevens (University of Strathclyde), with an account of the development of a nasal dosage form that can be used for the delivery of proteins and peptides and also, with obvious utility, nicotine.
Intranasal nicotine The paper was co-authored by the chairman, together with P. Thapa and A. J. Baillie (all from Strathclyde University, Glasgow). They described how lyophilisation was used to produce, from a range of concentrations of aqueous solutions of Methocels of various molecular weights, a stable powder containing nicotine hydrogen tartrate.
This was then tested in a diffusion cell with a minimal dissolution volume on the donor side to mimic the conditions in the nose, where there is a low-hydration environment. As would be expected, increased polymer concentration and viscosity reduced the rate of release of the drug below that obtainable from an aqueous solution, but the readily rehydratable and bioadhesive formulation was nevertheless very effective in vitro.
In vivo, it has the additional advantage that administration by the nasal route avoids first-pass metabolism and, as a concept, the new dosage form has merited a patent application.
Buccal nicotine tablet Many alternative controlled dosage forms of nicotine are available as aids to stopping smoking, one of which is the bioadhesive buccal tablet. O. Shakoor, D. F. Bain, N. C. L. Duguid and C. R. Park (Robert Gordon University, Aberdeen) have formulated such a tablet in the past, but this year they reported on the effects of adding chitosan to the formulation to improve both the adhesion and the control of nicotine release.
The substitution of chitosan for the carbopol of the original formulation resulted in almost perfect zero-order release, which is what is wanted. The bioadhesion was also improved, but increasing the chitosan percentage in the tablets produced excessive swelling in use, which was counteracted by the addition of polyvinyl pyrrolidone (PVP).
The optimal formulations, which are an improvement on earlier ones, contain nicotine 5mg, carbopol 1.5mg, chitosan 0.3mg, PVP 3mg, magnesium stearate 0.3mg and lactose 20mg.
Temperature and release rate Mixtures of a poly (d,l-lactide) Resomer R202H, of molecular weight around 10 kD, with an oligomeric version — R104 — that has a molecular weight of some 2 kD, display the remarkable property of having a release rate of drugs confined within them that varies greatly with temperature.
D. F. Bain (Robert Gordon University, Aberdeen), C. Rostron (John Moores University, Liverpool) and A. Smith (West Pharmaceutical Services, Nottingham) prepared microspheres from different blends of the two polymers. They then measured the release rates of rifampicin and of piroxicam, contained at the 20 per cent level, in a USP II paddle apparatus at 100 rpm and 33, 35, 37 and 39C (ie, temperatures straddling the physiological).
The polymer mixtures are inherently hydrophilic and their glass transition temperatures are controllable, being a function of the proportions of the two components. As they hydrate, matrix plasticisation occurs, together with a dramatic increase in release rate. This rate increase is itself a function of how close the glass transition temperature of the polymer mixture is to the temperature of the dissolution medium.
The authors made Arrhenius plots to allow activation energies for the transfer process to be determined. As might be expected, they decreased by almost a half as the proportion of the low molecular weight component was increased from 28 to 36 per cent.
The results give rise to the interesting possibility of increasing drug release rates after administration of a dosage form by the application of an extra-corporeal heat source.
Colon-specific delivery One method of enabling a dosage form to pass unchanged through the major portion of the gastrointestinal (GI) tract, and for drug release to occur only on reaching the colon, is to coat it with amylose, a high molecular weight component of starch, that is digested only in the colon by the activity of bacteria that are resident there and secrete amylase enzymes.
A. W. Basit, J. M. Newton (School of Pharmacy, University of London) and L. F. Lacey (Glaxo Wellcome, Greenford) have carried out a study to test out the efficacy of this administration procedure in humans.
The drug chosen was ranitidine. It was formulated into 1.5mm diameter pellets by extrusion-spheronisation, the pellets then being fluid-bed coated with a mixture of amylose (extracted from pea starch) and ethylcellulose, a water-insoluble polymer.
Preliminary experimentation showed that the best coating to provide resistance to upper GI tract conditions contained 25 per cent amylose and 75 per cent ethylcellulose, with the coating thickness adjusted to provide a weight gain of 20 per cent on the initial weight of the pellets.
Male human volunteers then received, on two separate occasions, 150mg of ranitidine as the coated formulation and an intravenous injection of 50mg of ranitidine solution. Technetium-99m radiolabelled pellets, identical in size and density to those containing ranitidine, were administered as tracers that were followed, as they traversed the gut, by a gamma camera. Blood samples were also taken, so that gut location could be correlated with drug absorption, and absolute bioavailability measured.
The results demonstrated that there was no ranitidine in the bloodstream until coated pellets had reached the colon, confirming colon-specific delivery. The absolute bioavailability was relatively low, at 5.6 per cent. This was not surprising, as it has been shown by the authors that ranitidine is metabolised under simulated colonic conditions, as they reported in another communication, an account of which follows.
A batch fermenter system was used to simulate conditions in the colon: faeces (10 per cent w/w) were homogenised in phosphate buffer at pH8 and 37C in three 100ml fermentation vessels, containing 0.1, 0.2 and 0.5g of ranitidine. They were sealed under nitrogen and placed in a rotating incubator to provide agitation, and sampled at intervals over 24 hours, the ranitidine content of the samples being determined by HPLC.
The ranitidine added in the two smaller quantities disappeared — the 0.1g amount over 12 hours and the 0.2g over 24 hours. The larger amount plateaued out at around 0.3g remaining. The colonic bacteria were able to metabolise the smaller amounts but were either saturated with, or killed by, the larger amount.
Intravaginal rings Silicone elastomeric rings are a useful controlled-release device for the intravaginal delivery of hydrophobic drugs, such as contraceptive steroids. Only hydrophobic drugs have a sufficiently high solubility in the silicone elastomer to maintain an effective dosage rate over several months.
In order to extend this delivery method to a wider range of drugs, R. Gallagher, R. K. Malcolm, J. Russell and A. D. Woolfson (Queen's University, Belfast) decided to introduce three common hydrophilic pharmaceutical materials into the silicone polymer, with a view to its use for the delivery of, for example, peptides. They introduced PVP, PEG and HEC into the polymer at levels of 5, 10 and 20 per cent, then measured the effect on the tensile and compressional properties of the resulting materials. PEG is a liquid, while the other two additives are powdered solids.
The tensile strength was considerably decreased by the former, which diluted the polymer network, but was decreased to a lesser extent by the latter, the effect in all cases increasing with increasing excipient concentration. On the other hand, PEG had little influence on the compressional properties, whereas the two solids acted as reinforcements, making the polymer more difficult to compress.
Similar results were obtained in flexural testing of polymer strips, where the solid additives gave an increased stiffness. After a week's soaking in water, the extraction of the hydrophilic additives left a microstructure consisting of a network of channels, the formation of which should allow any included drugs to be slowly released.
Intravenous isoflurane Isoflurane is an inhalation anaesthetic finding wide usage for both human and veterinary anaesthesia but, since it is an organic liquid, with a boiling-point of 48.5C, it cannot be administered intravenously, however convenient this might be. E. J. Gray, R. L. Horder and B. C. Withers (Abbott Laboratories, Queenborough, Kent) reported on their attempts to formulate a parenteral emulsion that could be injected to produce anaesthesia.
An emulsion was prepared in which the isoflurane was incorporated into soybean oil and an emulsion made using lecithin as the emulsifier. The suitability of formulations was assessed by droplet size analysis (laser diffraction, photon correlation spectroscopy), resistance to creaming on centrifugation and physical stability measurements.
A minimum of 30 per cent soybean to isoflurane in the disperse phase was necessary to attain stability, and the maximum overall isoflurane content was limited to 20 per cent v/v by viscosity considerations.
Tests on dogs showed that, at 0.75 ml/kg of a 10 per cent emulsion, anaesthesia was just beginning. At dosages of 1.0 and 1.25ml/kg, anaesthesia was rapid and effective, with smooth recovery after some 15 minutes. However, at a dosage of only 1.5ml/kg, anaesthesia was too deep, with cardiopulmonary collapse in two animals out of four. It had to be accepted, therefore, that the therapeutic window was too narrow for safety.

Medicinal chemistry: enzyme inhibition and pro-drugs

Three papers in this section, which was chaired by Professor D. E. Thurston (University of Portsmouth), described the syntheses of enzyme inhibitors that potentially may have useful clinical applications. The first was in the field of prostatic cancer.
Flavones as enzyme inhibitors Prostatic growth is dependent on dihydrotestosterone, produced from testosterone by the action of 5-a-steroid reductase, testosterone itself being derived from the weak androgen androstenedione by the action of 17-b-hydroxysteroid dehydrogenase.
R. Lelain, P. J. Nicholls, H. J. Smith and F. H. Maharlouie (University of Cardiff) have screened a large number of compounds that might act as inhibitors of this last enzyme, and reported that a number of flavone compounds possess the required inhibitory capability.
The experimental method was to incubate tritium-labelled androstenedione with a human microsomal testicular preparation, with and without added inhibitor, adding 14-C-labelled testosterone as an internal standard for subsequent analysis by TLC.
The most potent inhibitors were 7-hydroxy-flavone and baicalein, a trihydroxyflavone. Further work will be based upon the flavone skeleton as a promising starting point.
Coumate analogues As a treatment for hormone-dependent breast cancer, steroid sulphatase inhibitors are useful, since they block the action of oestrone sulphatase, which catalyses the hydrolysis of oestrone sulphate to oestrone, and also inhibit the action of dehydroepiandrostenedione sulphatase, which regulates the production of androstenediol, an oestrogenic steroid.
D. Ganeshapillai, L. W. L. Woo, B. V. L. Potter (University of Bath), A. Purohit and M. J. Reed (Imperial College School of Medicine) noted that 4-methylcoumarin-7-O-sulphamate (coumate) is an orally active, non-steroidal, site-directed inhibitor of both the sulphatase enzymes, as shown by tests on placental microsomes.
The authors therefore synthesised a number of coumate analogues, initially having alkyl chains of various lengths on carbons 3 and 4 of the coumarin ring, which turned out to have increased activity. The best results were obtained with a seven-carbon chain on carbon 3 and an eight-carbon chain on carbon 4.
The higher potencies were attributed to the fact that the bicyclic coumarin system mimicked the A and B rings of the steroid molecule. Therefore, it was reasoned that joining the alkyl chains to make a third ring that would be similar to the steroid C ring might increase potency still further.
Such was the case. The compound, with five carbons in the ring resulting from fusion of the two alkyl chains, proved to be active enough to go forward as a Phase I clinical trial candidate, and a compound with a seven-carbon ring has more recently been synthesised and has even greater activity.
Telomerase inhibition by anthraquinones The third presentation on inhibition had telomerase as its subject. This is an RNA-dependent DNA polymerase enzyme that elongates the 3Ē end of eukaryotic chromosomes. Most human cancers over-express this enzyme, so that it has become a target for anticancer chemotherapy.
V. Gibson, R. J. Anderson, D. Cairns (University of Sunderland) and T. C. Jenkins (University of Greenwich) have synthesised a number of anthraquinone analogues and tested them for telomerase inhibitory capability.
The compounds were readily obtained, in good yield, by refluxing the appropriate chloro-substituted anthraquinone with an excess of either NĒN-diethylamino-ethylenediamine or 5-amino-pentanol.
All the synthesised compounds inhibited telomerase at a concentration of 10 micromolar. None of them inhibited the related enzyme — Taq polymerase — even at concentrations as high as 50 micromolar. Most importantly, none showed any evidence of general cell toxicity in tests using ovarian cancer cell lines.
Telomerase inhibition Commenting that telomerase has been found to be active in over 90 per cent of human tumours, M. Wan, P. Fell, P. Z. Sayyed and S. Akhtar (University of Aston) investigated the use of catalytic RNA and DNA molecules as telomerase inhibitors.
They synthesised RNA and DNA enzymes, using an ABI 392 DNA/RNA synthesiser, that were designed to target the RNA component of human telomerase. The activity of the enzyme, and its inhibition, were measured in tumour cell lysates using a telomeric repeat amplification protocol.
The hammerhead ribozyme that was finally designed had a dose-dependent capability for telomerase inhibition in human glioma cell lysates with an IC50 of 0.4 micromolar. Such nucleic acid enzymes are showing great promise in tumour treatment.
Temozolomide pro-drug The anticancer drug temozolomide has been used against brain tumours. It is poorly soluble in both aqueous and organic solvents, but has good oral bioavailability and can cross the blood-brain barrier. It also shows activity against leukaemia, breast cancer and melanoma.
D. L. Rathbone, D. Su, Y. Wang and D. C. Billington (University of Aston) reported on how they wished to formulate a water-soluble pro-drug form of both temozolomide and the closely related mitozolomide that would enable the drugs to be injected, and would direct them elsewhere in the body than the brain. Accordingly, they converted the compounds into the corresponding carboxylic acids, and then esterified these with a range of mono- and bi-functional polyethylene glycols.
The basis of the authors' thinking was that the polymer would impart water solubility and prevent migration across the blood-brain barrier. Esterase activity in the body would regenerate the carboxylic acid form plus the non-toxic polyethylene glycol, and, since the carboxylic acid is ionised at physiological pH, it too could not cross the blood-brain barrier.
That this actually occurred was demonstrated by exposure of the preparation to pig liver esterase, and by following the course of the hydrolysis reaction by NMR spectroscopy in deuterated phosphate buffer.

Physical pharmaceutics in 1999

Professor G. Buckton (School of Pharmacy, University of London) chaired this session, which started with a keynote address by M. E. Aulton (De Montford University, Leicester) on the topic of the manipulation of the properties of crystalline materials in order to assist the dosage form designer.
For tablet production, crystals should be grown rapidly, in order that there should be the maximum number of imperfections in the lattice due to the lack of time for molecules to align themselves correctly at the growing crystal surface. Large numbers of imperfections imply weaker crystals that will deform more readily and consolidate better. For dry powder inhalation products, the grown crystal must either be relatively brittle, so that it can be milled to the required small particle size, or it must be grown only to a small size in the first place. This is an application of the emerging technology of dissolution in, and precipitation from, supercritical fluids. Also, since the small inhalation particles must be carried on larger ones to obtain a non-cohesive formulation, control of the surface adhesional capability of the crystals must be achieved. This is done by manipulation of the solvent system, to ensure suitably weak adhesion to the carrier particles, so that separation will occur under the flow conditions prevailing in the inhalation device.
Surface free energy The keynote lecturer also presented the results of some measurements of tablet surface free energies, obtained with co-authors K. J. Lennon, A. M. Twitchell (De Montfort University, Leicester), A. J. Coupe and S. R. Brown (Pfizer Research, Sandwich, Kent).
The molecules at the surface of a solid possess surface free energy that can be assessed by measuring the contact angle that a liquid, or preferably several different liquids, make with the tablet surface. Model tablet formulations were prepared from Avicel PH102 (32.75 per cent w/w), Ac-Di-Sol (2 per cent), lactose (around 60 per cent) and magnesium stearate at a level of 0.25, 2 and 5 per cent. The amount of lactose was adjusted in each formulation to accommodate the added magnesium stearate, while keeping the percentages of the other ingredients constant.
A second series of similar formulations was prepared, with dibasic calcium phosphate replacing lactose. Two blending regimes were used for each formulation. In the first, all the components were taken at once and blended for 10 minutes. In the second, everything except the magnesium stearate was blended for 10 minutes, at which point the stearate was added and blended in for a further 60 seconds. These conditions simulated under- and over-lubrication, respectively. Tablets of diameter 12.7mm were then made under 10kN compaction force and contact angles measured, using a sessile drop technique, between the tablets and water, di-iodomethane or ethylene glycol. In general, increasing the quantity of magnesium stearate and decreasing the time for which it was blended in, increased the measured surface free energy, the effect being the same for both lactose and calcium phosphate tablets. Such surface free energies may be used to predict the likely degree of spreading of film-coating solutions over tablet surfaces, and their adhesion.
Testing droplet size Inhalation formulations should produce a high proportion of particles or droplets having diameters between 1 and 5mm — the so-called respirable fraction. To assess this in vitro, various test devices (artificial lungs) have been used.
An improved version is being developed by R. Hopkins, M. J. Tobyn, J. N. Staniforth (University of Bath), I. T. Shrubb and P. Wright (Astra Charnwood, Loughborough). It is called the shell model and is a series of five hemispherical shells, four of which contain holes drilled in a regular pattern, representative of the airway spaces in the relevant part of the lung. The fifth, the final stage, is connected to the actuating vacuum pump through a filter.
The performance of the device was tested on three metered dose inhaler formulations — Ventolin, which contains a CFC propellant, Ventolin Evohaler and Airomir, both of which use HFA propellants. The results show that particles reached the various shells in the manner expected. The next stage of the work is to compare the device with in vivo data such as can be obtained by, for example, gamma scintigraphy.
Dosage delivered by inhalers Metered dose inhalers (MDIs) should produce a constant amount of active ingredient over the whole of their life (some 200 actuations). However, it has been predicted, on theoretical grounds, that usage of the propellant could result in a gradual increase in drug concentration in the remaining liquid phase, so that the later doses are larger than the earlier.
D. Lewis, D. Ganderton, B. J. Meakin (University of Bath), G. Branbilla (Chiesi Farmaceutici, Parma, Italy) and D. Howlett (Bespak, King's Lynn) tested out this prediction by discharging actuations 6 and 200 of two different HFA-propelled flunisolide inhalation formulations from MDIs into a dose unit sampling apparatus, then washing out, collecting and assaying by HPLC the delivered dose (from the sampler) and the metered dose (sampler plus drug deposited on the actuator).
In all cases, the 200th delivery was 10 per cent larger than the sixth, as predicted. This difference was felt to be sufficiently significant to merit consideration in any revision of the 1998 FDA Guidelines for MDIs, which currently require a dose uniformity of within 15 per cent of the label claim, and make no allowance for changes of the type reported here that occur during the can lifetime.
Water geometry On a completely different tack, the properties of bound water are of scientific interest because rings of water molecules are known to exist near the surfaces of proteins, at DNA interfaces and in the hydration shells of DNA-drug complexes. Such water will probably have much in common with that in clathrate hydrates.
Being interested in the computer simulation of the interaction of water and drug molecules, T. H. Plumridge, R. D. Waigh (University of Strathclyde) and G. Steele (Astra Charnwood, Loughborough) carried out a survey of the water geometries found in the known hydration structures of larger biological molecules, and also of carbohydrates and small components of nucleic acids. They aimed to get a distribution of the values of bond angles and interatomic distances for the assembling of water molecules by hydrogen bond formation.
The values were then introduced as constraints into a computer simulation that builds networks of water molecules by successive molecular introduction. At each stage, the water chain is rotated through all its allowable hydrogen bond configurations, and the next molecule is attached when the angle and distance satisfy the constraints. At each addition, therefore, the ice structure is maintained.
It is hoped to be able to simulate the hydration of a variety of compounds of pharmaceutical interest.

Developments in pharmaceutical analysis

This session, held on the Tuesday afternoon and chaired by Professor A. C. Moffat (School of Pharmacy, University of London) started with a keynote address on trends in liquid chromatography, given by R. M. Smith (University of Loughborough).
Over the past few years, there has been a gradual increase in the use of liquid chromatography, coupled with mass spectrometry, as a routine technique. With the increased usage has come a decrease in unit costs, as smaller and more user-friendly instruments have appeared. Detectors and stationary phases have made some, but not spectacular, progress.
The most likely future advances would appear to be in the field of electrically driven methods, such as capillary electrochromatography, which is slowly overcoming its operational problems. Such methods are the most likely to result in the ideal of "chemistry on a chip".
New technologies forum The first presentation took the unusual form of a description of the New Technologies Forum that has been set up by the Royal Pharmaceutical Society with the intention of spreading the awareness of new technologies and their probable impact on both the pharmaceutical industry and regulatory affairs.
It held its first meeting last February, when the topic was Raman spectroscopy. Regulators present at that meeting confirmed that its use had rarely been registered, and inspectors confirmed they had not observed Raman equipment in use in the UK, though obviously it is widely used in research and development departments. Future forum topics include acoustic emission spectroscopy, statistics, and process measurement and control.
Atomic force microscopy Atomic force microscopy is a recently developed technique, in which a probe can be tracked over a surface to follow its structural arrangement at the nanometre scale of scrutiny. The technique has great potential for looking at biological systems, such as proteinaceous layers. M. M. Stevens, S. Allen, W. C. Chan, M. C. Davies, C. J. Roberts, S. J. B. Tendler and P. M. Williams (University of Nottingham) described its use in investigating the molecular architecture of a complex bilayer.
On a silicon substrate, they constructed a layer of streptavidin, overlaid with a layer of a bis-biotinylated peptide. On this was superposed another streptavidin layer to form a sandwich, held together by the high affinity interaction between biotin and streptavidin. The construction of the layer has been monitored in real time, with both its surface affinity and mechanical properties being followed. In particular, as it grew, there was an increase in its stretchability. The high affinity between the two major components of the film have led to its being used in a number of immunochemical and diagnostic assays.
Near-infrared spectroscopy The session chairman has, over the past few years, brought into prominence the useful technique of near-infrared (IR) spectroscopy. This year, he co-authored eight papers describing analyses that could be carried out using near-IR methods. Space precludes a detailed reportage of each, but it seems informative to list them briefly.
They were: (1) an improvement in a previously reported method of measuring the particle size of a powder, such as microcrystalline cellulose; (2) a method of following the course of a chemical reaction (the esterification of a resin-bound alcohol) in real time; (3) the rapid identification of some essential oils; (4) the determination of the water content of a pharmaceutical preparation through the amber glass wall of the container; (5) the quantification of codeine present as a minor constituent in intact tablets; (6) a similar assay of the active ingredient, a development compound, within intact capsules; (7) the differentiation of five digitalis species; and (8) a feasibility study on the setting up of an inter-institutional library of near-IR spectra for rapid reference purposes.

Short papers
There was a second pharmaceutical analysis science session, of so-called short papers, on the Wednesday morning, in association with the Joint Pharmaceutical Analysis Group, chaired by Dr H. M. Hill (Huntingdon Research Centre).
Bioglass analysis This was an account of the analysis of Bioglass, a bioactive ceramic material that is capable of bonding with bone and soft tissue to improve healing after injury. T. Fairbrother, R. Almond, J. Tse, I. Way and A. Roberts-MacIntosh (Smithkline Beecham, Weybridge) reported on the use of diffuse reflectance Fourier transform spectroscopy and inductively coupled plasma spectroscopy in the characterisation of the material, which contains oxides of silicon, sodium, calcium and phosphorus. It depends for its success on the formation of a surface layer of a hydroxyl-carbonate-apatite compound as very small crystallites.
The layer forms in five stages when immersed in an aqueous medium. There is a rapid exchange of ions with the medium, resulting in a slight loss of soluble silica and a surface depletion of alkaline earth metal ions, with the formation of silanols on the glass surface. Calcium and phosphate ions then migrate through the silica-rich surface formed by re-polymerisation of the silanols, and pick up hydroxyl ions to form the required apatite layer. All these stages can be followed by the two spectroscopic techniques used by the authors.
Chiral separation of benzodiazepines Short-acting benzodiazepines with a 3-hydroxy group are chiral molecules. Temazepam and oxazepam are metabolites of diazepam, but it is not currently known whether the conversion is stereospecific. A chiral separatory method was sought by D. G. Durham, A. Lee and A. S. Low (Robert Gordon University, Aberdeen), who initially tried adding b-cyclodextrin to the mobile phase, but without achieving a satisfactory separation. Success was finally obtained using a 5mm Hypersil-packed column 200mm long and 4.6mm diameter, in reversed phase with acetonitrile or methanol-phosphate buffer as solvents, and hepta-kis-(2,6-dimethyl)-b-cyclodextrin as a mobile phase additive. This success, which was only partial since lorazepam was not resolved, was ascribed to the fact that the molecule of the substituted cyclodextrin has a larger cavity and so forms a more stable inclusion complex.
Water content of pharmaceuticals The most common method for the estimation of the water content of pharmaceuticals is, of course, the Karl Fischer method, but it has its limitations. Some drugs are not soluble in methanol, the solvent used, and some drugs have functional groups that can interfere with the determination. S. A. Tabatabai, A. Shafaati and A. Asgari (Shaheed Beheshti University, Tehran) reported their development of a new method that overcomes these disadvantages.
Water in the specimen to be assayed is selectively derivatised with phenyl isocyanate to produce, quantitatively, N-NĒ-diphenylurea, which can be separated from interfering substances and quantified by HPLC. The separation was performed on a 300 by 3.9mm Bondapack C18 column with a mobile phase consisting of 50:50 methanol and water. The method was tested on ampicillin, camphor, cephalexin, ibuprofen, methyldopa, oxytetracycline, piroxicam and povidone, on all of which it succeeded, with a lower limit of detection of 5mg of water per ml of sample.
Fungus identification The time-of-flight mass spectrometry already mentioned as being a useful new technique was used to identify fungi by K. J. Welham, M. A. Domin, D. S. Ashton, K. Johnson and R. K. S. Ulirs (School of Pharmacy, University of London). The method has been used previously for the identification of bacteria. It is based on the application of matrix-assisted laser desorption/ionisation mass spectrometry and electro-spray mass spectrometry, which enable high molecular weight proteins, oligosaccharides and oligonucleotides to be analysed because large ion fragments are produced.
Fungal cells are larger than those of bacteria and are 80-90 per cent polysaccharide (including chitin for strength). Because of this high carbohydrate content, analysis is best performed in negative ion mode, which is well recognised to be more sensitive for carbohydrate samples. The authors have applied the method to Penicillium species, Scytalidium and Trichophyton, each showing a distinctive spectrum over the mass range 2 to 13 kDa. Because, as an essential part of sample preparation, washing in methanol is required, which may selectively release components from the fungal cell wall, further investigation will be needed to optimise the analytical protocol.
Spectrophotometry of antifungals Moving from fungi to antifungal drugs and their estimation, a spectrometric and a spectrofluorimetric analytical method for the assay of the antifungals clotrimazole, ketoconazole, miconazole, econazole nitrate and tolnaftate has been developed by P. Y. Khshaba, S. R. El-Shabouri, K. M. Emara and A. Mohamed (Assuit University, Egypt).
The spectrophotometry was based upon the formation of a complex between the electron-donating tertiary amine moiety of the antifungal and one of the acceptor reagents. These were 2,3-dichloro-5,6-dicyano-1,4-benzoquinone in methanol solution or p-chloranilic acid in acetonitrile. The coloured complexes in solution obeyed Beer's law over a wide range and so could be quantified by absorbance measurements, the first type at 450nm and the second at 520nm.
Ketoconazole has intrinsic fluorescence, so that analysis by spectrofluorimetry could be carried out using the light emitted at 375nm as a consequence of excitation by light of wavelength 288nm.
The analysis of tolnaftate was not so straightforward. It was dissolved in methanol and quantitatively hydrolysed with 5 molar sodium hydroxide solution: the hydrolysate, when excited with light of wavelength 344nm, fluoresced at 420nm. The new methods were applied to all the drugs in their various formulations — powders, creams, tablets, suppositories, topical solutions and oral gels.
In all cases, results were in agreement with official methods.

Pharmacognosy: sunscreens, HIV and cytotoxics

On the Wednesday afternoon, the session chairman was Professor P. J. Houghton (King's College London). He introduced R.Verpoorte (University of Leiden, The Netherlands), who gave the keynote lecture on the role of biotechnology in pharmacognosy.
Plant cell cultures can be used for the production of pharmaceuticals, although in most cases productivity is very low, and the main usefulness of such methods is ensuring a small, continuous, but extremely valuable supply of compounds during the early stages of their development. Metabolic engineering is a method of increasing the production rate by plants of compounds of interest. There are several pursuable strategies — increasing the activity of rate-limiting enzymes, blocking pathways competitive with the desired one, or introducing heterologous genes to encode a different metabolic branch altogether.
In the speaker's laboratory, genes encoding for tryptophan decarboxylase and strictosidine synthetase have been obtained from Catharanthus roseus, and introduced into a variety of plant cells. In yeast, feeding with tryptamine and secologanin then results in the production of large amounts of indole alkaloids. Similar methods have resulted in plants with the ability to synthesise salicylic acid, an important signal compound in the acquirement of systemic resistance. A decrease in the amount of necrosis observed after infection with viruses was correlated with the increase in salicylic acid content.
Screening of natural materials Of the 20 top-selling pharmaceuticals in the decade from 1985 to 1995, around half had active ingredients with a chemical structure based upon a compound originally found in nature. Despite the increased potential of robotic synthetic screening techniques, a natural products programme is still worth maintaining.
J. Lewis (Glaxo Wellcome R&D, Stevenage) described his company's policy for the acquisition and screening of natural-source materials, noting particularly that there is a large amount published on the uses of plants in traditional medicine, and that there exist data bases, for example at the University of Illinois at Chicago, containing reports on biological activity, natural products and ethnobotanical literature.
As an example of the general method, the author quoted the discovery of some euphane triterpenes. In a search for inhibitors of human thrombin (of value in the treatment of deep vein thrombosis), 150,000 samples were screened. They included bacterial, fungal and plant extracts, and also synthetics. The result was that a methanolic extract of Lantana camara (Verbenaceae), taken from a UK garden, showed considerable activity. After large scale extraction and bio-assay guided fractionation, it was established that the active constituent was a 5,5-trans-fused lactone, containing the triterpene, resulting in compounds showing IC50 values of 50 nanomolar against thrombin.
Medicinal plants for HIV Any compound that shows anti-HIV activity will be worthy of further investigation. The session chairman, together with T. Z. Woldemariam, (King's College London), S. O'Shea, J. E. Mullen, T. Rostron, J. E. Banatvala (UMDS, London), E. Walker (Ruchill Hospital, Glasgow), S. P. Thyagarajan, S. Soloman (University of Madras, India) and L. Y. Foo (Gracefield Research Centre, Lower Hutt, New Zealand), has examined the Indian medicinal plant, Phyllanthus amarus, that is traditionally used in the treatment of jaundice and liver disorders, and is active against the hepatitis B virus.
Extracts of the plant are already known to be able to inhibit reverse transcriptase, but activity in vitro against HIV-infected cells has not, so far, been investigated. Aqueous extracts of leaves, and also a number of solutions of galloyl ester polyphenolic compounds that have been reported from the plant previously, were tested for their ability to inhibit infectivity of a T-cell lymphoma preparation (cell line H9) when challenged with patient-derived HIV-1.
After a seven-day incubation, infection was assessed by the levels of p24 antigen that had been produced, and also by determining the highest virus dilution in which syncytia were observed.
The polyphenols amarulone, amariin, elaeocarpusin, corilagin and geraniin, and the aqueous extract itself, all showed a degree of anti-HIV activity sufficient to justify the reputation that the plant has acquired.
Mode of action of insecticidal oils There have been conference reports of the insecticidal activity of essential oils in recent years but not on their mode of action. C. M. Priestley, E. M. Williamson (School of Pharmacy, University of London) and D. B. Sattelle (University of Cambridge) have investigated the effect of three monoterpenoids, derived from the essential oils of three plants, on the GABA receptors of Drosophila melanogaster.
The test compounds were linalool, found in a number of oils, terpinen-4-ol, an alicyclic alcohol derived from tea tree oil, and thymol, an aromatic alcohol occurring, unsurprisingly, in thyme. The receptors were not in the flies, but were created by the heterologous expression of RDL GABA receptors. They were produced by the injection of the appropriate encoding cRNA into the cytoplasm of frog oocytes, followed by voltage-clamp electrophysiological measurement (the same technique as was used by Daniels, Wittmann and Fraser, as reported in the neuropharmacology session).
The amplitudes of the currents observed in response to a 30-second period of stimulation by a GABA concentration of 10-5 molar were increased by 148 per cent by a two-minute pretreatment with the same molar concentration of thymol, though unfortunately the other two terpenoids had no effect. However, the reversible and dose-dependent response indicates that thymol is a new allosteric modulator for insect GABA receptors and merits investigation as an insecticide.
Sunflower sunscreen Another useful essential oil is that of the sunflower seed, Helianthus annuus, which has a high UV absorbance and is thought to enhance melanin production in the skin.
E. Farboud, G. H. Amin and M. Fazollahi (University of Tehran, Iran) extracted the oil by the Soxhlet technique, using ether as the solvent, and then formulated a water/oil emulsion containing the oil, lanolin, isopropyl myristate, beeswax and borax. They also made a cream containing 50 per cent of the oil in a polysorbate 80/sorbitan monostearate base. The UV absorption was measured, and the cream shown to have a sun protection factor of 5.
Sunscreen penetration vehicle dependent Sunflower oil is the only UV-absorbent material present in the above formulation, and all the ingredients are known to be innocuous. This is not the case with many commercial preparations, as was exemplified in another communication, from H. A. E. Benson, M. S. Roberts (University of Manitoba, Canada) and M. L. Nocente (University of Queensland, Australia).
These authors noted that a typical sunscreen formulation will contain a physical blocker, such as titanium dioxide, together with one or more chemical absorbers, of which benzophenone-3 is a widely used example. It can penetrate the skin, which is not desirable, and the penetration rate is dependent on the vehicle. The authors measured the penetration rate of the compound through human skin in a number of formulations.
Dissected abdominal skin was mounted in a Franz-type diffusion cell, and a range of formulations, each containing 2 per cent of benzophenone-3, was applied on the donor side. The acceptor side contents were monitored by sampling as a function of time and analysing for benzophenone-3 by HPLC.
The vehicles examined were light liquid paraffin, coconut oil, a 50:50 mixture of coconut oil and ethanol, aqueous cream BP and oily cream BP. There were differences in penetration rate and total amount absorbed, but the most important conclusion was that alcohol-based formulations should be avoided.
Algal anti-tumour compounds Anti-tumour compounds emerge from the most unlikely places: Y. Ezure, M. Ishida and T. Suzuki (Institute of Technology, Tokyo) are on the track of some that are extractable from the blue-green fresh water alga Anaboena.
The authors have previously made and tested extracts from A cylindrica and two strains of A variabilis. All were grown photoautotrophically in the laboratory under controlled conditions, and extracts were tested for their activity against mouse leukaemia cells. Aqueous extracts showed no activity, ethanolic extracts showed moderate activity, while chloroform extracts gave rise to some degree of optimism.
Following on from this, the authors reported that they have now separated the chloroform extract into 21 fractions by silica gel column chromatography, eluting stepwise with hexane/chloroform/methanol, monitoring the fractions by the cytotoxic assay and by TLC.
The dose-response curve for the most active fraction compares well with that for 5-fluorouracil, the MIC being only twice that of the control drug. However, examination by NMR, infrared and mass spectroscopy have shown that the extract is still a mixture of compounds, and no definitive structures have yet been assigned: work is still in progress.

Dosage form characterisation

Chaired by Professor D. Q. M. Craig (Queen's University, Belfast), the session began with a keynote address on the topic of protein freeze-drying, given by M. J. Pikal (University of Connecticut, US), who pointed out that the freeze-drying process is subject to more variability than might be expected.
Proteins can change their conformation during the process, and, as an example, the stability of human growth hormone in the frozen glassy state can vary by as much as two orders of magnitude, depending upon specific excipient effects. Molecular mobility is the important factor governing stability, and the measure of this is the structural relaxation time.
The author presented a fairly complicated relationship that described the variation of this parameter with temperature. Its value has been calculated for sucrose and indomethacin, with good agreement with experimentally determined values, and the relative stabilities of the sodium salts of two freeze-dried cephalosporins (cephamandole and cephalothin) were in almost exact proportion to their structural relaxation times.
Measuring lipid films A new method of measuring the properties of lipid films is the thickness shear mode sensor, which is an oscillating quartz crystal, the frequency of which will shift when it is loaded with an adherent film. Since frequency is a quantity that is very precisely measurable, film parameters that could not be accessed in any other way become determinable.
M. Reason, J. Ramsden, G. Smith, P. Teesdale-Spittle, R. Latham (De Montfort University, Leicester) and B. Henry (Pfizer Research, Sandwich, Kent) used such a sensor to study three fatty acid films: stearic (18 carbon chain), arachidic (20) and behenic (22). They were coated onto the sensor using a Langmuir-Blodgett apparatus, adding a one-molecule thick layer at a time, so that films of gradually increasing thickness could be built up. The frequency shift of the sensor increased rapidly and non-linearly as the first three layers were applied, then steadied off to a linear rise with layer number up to the maximum studied, which was a film 13 molecules thick. The frequency shift was smaller, the longer the fatty acid chain.
The initial curvature indicates that the films have a degree of viscoelasticity, but, once the linear portion has been reached, the frequency shift can be used to measure film thickness.
Examining artificial membranes A recently developed liquid chromatographic phase consists of phospholipids covalently bonded to an aminopropyl silica surface. The idea is to mimic half of a biological membrane bilayer. The system is called an immobilised artificial membrane. A closely related system can be prepared by dynamically coating phospholipid onto reversed-phase C8 columns. Neutron reflection studies to examine these layers, and to give a structure-related explanation of the chromatographic data obtained from them, have been carried out by C. M. Hollinshead, M. Hanna, D. J. Barlow, A. J. Hutt, M. J. Lawrence (King's College London), V. de Biasi, P. Camilleri (Smithkline Beecham, Harlow), J. Lu, T. J. Su (University of Surrey) and D. G. Bucknall (Rutherford Appleton Laboratory, Chilton, Oxfordshire).
The polished (111) surface of a silicon block was first treated with deuterated octadecyl-trichlorosilane, and the resulting surface layer characterised by neutron reflectivity, being immersed in a 2:3 mixture of ordinary and heavy water to give a contrast match with silicon. The pretreated block was then coated with dimyristoyl-phosphatidylcholine using a Langmuir-Blodgett apparatus and further examined by neutron reflectivity.
The data were fitted to a four-layer model — the first layer was 73 per cent silicon and 27 per cent silicon oxide, the second was 57 per cent trichlorosilane and 43 per cent phosphatidylcholine, the third was 51 per cent chlorosilane, 21 per cent phosphatidylcholine and 28 per cent water, and the fourth was 67 per cent phosphatidylcholine and 33 per cent water. The fourth layer contains the phospholipid head groups, while in the second and third layers the alkyl chains of the silane and the lipid are mixed.
The conclusion is that the coating should not be considered to be a supported half bilayer, since there is such a large degree of interpenetration of the constituents.
Artificial neural networks R. C. Rowe ( Astra Zeneca) has reported in the past few years on the computerisation of tablet manufacture and coating formulation. This year, with A. P. Plumb, C. Doherty (Astra Zeneca, Loughborough) and P. York (University of Bradford), he described some of the possible difficulties that could arise from the overtraining of artificial neural networks in coating formulation.
There were six input parameters (polymer molecular weight, film thickness, pigment type, pigment size, size distribution and concentration) and two output parameters (film opacity and film crack velocity) that had already been calculated by computer simulation.
A system containing up to nine nodes was trained with a data set of 102 records and validated using a composite designed data set of 80 records. There is a danger that after a few thousand training epochs, the system has, in essence, memorised the data set and, therefore, has limited its own ability to generalise, although performing excellently within its own (limited) scope.
Attenuated training is a method of avoiding this effect, in which the onset of overtraining is detected, and training stopped, when the global minimum of the associated test error is found, the system being allowed to consult an independent test data set.
It turns out that such attenuated training produces a system with lower, but acceptable, error levels that is more predictive than the fully trained equivalent.

Dr Ridgway is a private consultant in powder technology. Formerly, he was senior lecturer for 20 years at the School of Pharmacy, University of London. He took early retirement and has since become a psychologist